09-04-2022, 06:17 PM
(09-04-2022, 09:39 AM)Ofnuts Wrote: After a lot of head scratching, I still don't know how it got there, but I know why things are amiss: the internal representation of the pixel values in sunlite isn't a valid representation, which is what nan (Not A Number) means in the Pointer dialog.
If I try to alter the layer by painting on it, the bad part seems to be a rectangle, so was the layer smaller at some point and later enlarged (Layer to image size or else)?.
Also, are you using the OpenCL acceleration?
In the Libre suite of tools I'm familiar with where to find an OpenCL setting. I looked through my Win11 install of GIMP2.10's Preferences and could not find a tickbox to the effect. I'm attaching settings screenshots which I think might be relevant.
I cannot recall if I resized the 'sunlite' layer, though I will be more vigilant in tracing any exact changes to a layer I've made before the black Nothing kicks in.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Quote:How to get rid of them and keep the contents of the layer ? So far this should work. Clear the sunlite layer (or a selection of the affected area). Make a new empty transparent layer under the 'sunlite' layer. Merge the 'sunlite' layer down into the new empty transparent layer.Thank you Rich, I will try this!
1 minute example: https://i.imgur.com/1y9GfkJ.mp4
You may have saved me once again...
On my Linux Mint appimage install of 2.10, there are sometimes strange effects that can emerge from GMIC (I doubt you remember this Rich, but you turned me on to GMIC several years ago by teaching me about the Extract Objects filter!) or even some default filters. Oddly enough, they aren't black nothing clouds, but brightly random pixels (as if you applied a Hard Mix mode to the layer or something). Like the black nothing the chromatic pixels don't go away and spread if you keep working the layer. I just find it interesting that the behavior of the "bug" is the same but the qualities of it so different.